fMRI neurofeedback of higher visual areas and perceptual biases

Abstract

The self-regulation of brain activation via neurofeedback training offers a method to study the relationship between brain areas and perception in a more direct manner than the conventional mapping of brain responses to different types of stimuli. The current proof-of-concept study aimed to demonstrate that healthy volunteers can self-regulate activity in the parahippocampal place area (PPA) over the fusiform face area (FFA). Both areas are involved in higher order visual processing and are activated during the imagery of scenes and faces respectively. Participants (N=9) were required to upregulate PPA relative to FFA activity, and all succeeded at the task, with imagery of scenes being the most commonly reported mental strategy. A control group (N=8) underwent the same imagery and testing procedure, albeit without neurofeedback, in a mock MR scanner to account for any non-specific training effects. The upregulation of PPA activity occurred concurrently with activation of prefrontal and parietal areas, which have been associated with ideation and mental image generation. We tested whether successful upregulation of the PPA relative to FFA had consequences on perception by assessing bistable perception of faces and houses in a binocular rivalry task (before and after the scanning sessions) and categorisation of faces and scenes presented in transparent composite images (during scanning, interleaved with the self-regulation blocks). Contrary to our expectations, upregulation of the PPA did not alter the duration of face or house perception in the rivalry task and response speed and accuracy in the categorisation task. This conclusion was supported by the results of another control experiment (N=10 healthy participants) that involved intensive exposure to category-specific stimuli and did not show any behavioural or perceptual changes. We conclude that differential self-regulation of higher visual areas can be achieved, but that perceptual biases under conditions of stimulus rivalry are relatively robust against such internal modulation of localised brain activity. This study sets the basis for future investigations of perceptual and behavioural consequences of localised self-regulation of neural activity.

Keywords: Binocular rivalry; Fusiform face area (FFA); Neurofeedback; Parahippocampal place area (PPA); Self-regulation.

Fig. 1

The components of the neurofeedback setup. Data are acquired in the MRI scanner and transferred in real time to an analysis computer, where they are also analysed in real time (for example one frame every 2 s). A translation algorithm then converts the extracted feature of the MRI signal into a command for a feedback signal, which can be displayed to the participant while he or she is in the scanner.

Fig. 2

Task design. (A) Binocular rivalry task: Each block consisted of four trials that lasted 100 s each. All trials were separated by a 30 s rest period and both blocks were separated by a rest period of 110 s. Participant are presented with an image of a face superimposed on an image of a scene and indicate via button presses whether the face or scene is experienced as more pronounced. (B) Neurofeedback run with the judgement task: Each run consists of 3 cycles of up-regulation, judgment task (with either gender or indoor/outdoor judgements) and rest.

Fig. 3

Self-regulation of PPA vs. FFA. (A) Activation levels in the parahippocampal place area (PPA) and fusiform face area (FFA) in the three scanning conditions (localiser, judgement task, self-regulation), represented by the t-statistic of the respective predictor and averaged over all runs. As expected, real-stimuli resulted in stronger activation than imagined stimuli. (B) Individual data for the nine participants. These show that, regardless of interindividual variability in activation levels, participants were generally able to obtain a higher activation level in the PPA (orange bars) than in the FFA (dark blue bars) during the self-regulation condition, as per instruction.

Fig. 4

Localisation of the target areas. Results from an individual localiser run, showing a contrast map for face – (in cold colours) vs. scene – (in hot colours) related activation. Target areas selected for neurofeedback in the right FFA (purple) and left PPA (green) are indicated. Image presented in radiological convention (right side of the brain is left side of the image).

Fig. 5

Activation patterns in five areas activated during both the localiser (LOC) run and neurofeedback (NF) run. Activation levels (as ratio of PPA activity) are shown for the Middle frontal gyrus (MFG), Frontal eye fields (FEF), Superior parietal lobule (SPL) and the Visual cortex (V1). Areas with a larger synaptic distance from the primary visual cortex (V1) seem to be activated relatively more during the neurofeedback than the localiser runs.